Jackson data fiddlings
Misty
2025-02-21
G-test
att = read.csv("attitudes.csv", header = T)distance v attitude for SEOT
From qualtrics: There is a statistically significant relationship between Q8_1: When you encounter sea otters and harbor…lose do you generally get to them? | sea otters and Q9_4: What is your general attitude towards sea otters and harbor seals? | sea otters
Q8v9 = att %>% select ("Q8_1", "Q9_1", "ResponseId")
dim(Q8v9)## [1] 331 3turn Q8_1 into an integer
Q8v9$Q8_1 = as.integer(Q8v9$Q8_1) bin distances
Q8v9$binned.Q8 <- as.factor(ifelse(Q8v9$Q8 <=10, 'B1 <=10',
ifelse(Q8v9$Q8<52, 'B2 11-51',
ifelse(Q8v9$Q8<103, 'B3 52-102',
ifelse(Q8v9$Q8<154, 'B4 103-153',
ifelse(Q8v9$Q8<205, 'B5 154-204',
ifelse(Q8v9$Q8<256, 'B6 206-255',
ifelse(Q8v9$Q8 <300, 'B7 256-299',
ifelse(Q8v9$Q8 >=300, 'B8 >= 300', 'Transfer')))))))))remove any rows with NAs
Q8v9 <- na.omit(Q8v9)
dim(Q8v9)## [1] 289 4Response ID R_12KtQUffRaYUb7b still had no value for q9 so I omitted that.
Q8v9 = Q8v9[!grepl("R_12KtQUffRaYUb7b", Q8v9$ResponseId),]
dim(Q8v9)## [1] 288 4Create a contingency table (I made sure it matches what was in Qualtrics)
Q8v9_table = table(Q8v9$binned.Q8, Q8v9$Q9)
#Q8v9_tableRun a Chi-square test for independence; these results are the same as what came from Qualtrics
chisq_Q8v9 = chisq.test(Q8v9_table)
chisq_Q8v9##
## Pearson's Chi-squared test
##
## data: Q8v9_table
## X-squared = 38.543, df = 14, p-value = 0.0004289Look at which pairs contribute most to the significance. First look at the Pearson’s residuals
round(chisq_Q8v9$residuals, 2)##
## I don't like seeing them I love seeing them
## B1 <=10 -0.50 -0.69
## B2 11-51 -0.49 -0.31
## B3 52-102 0.31 0.27
## B4 103-153 -1.05 0.42
## B5 154-204 -0.58 0.23
## B6 206-255 1.40 -0.08
## B7 256-299 2.78 -0.33
## B8 >= 300 0.35 -0.02
##
## Neither like nor dislike
## B1 <=10 4.00
## B2 11-51 2.00
## B3 52-102 -1.64
## B4 103-153 -1.36
## B5 154-204 -0.75
## B6 206-255 -0.67
## B7 256-299 -0.42
## B8 >= 300 -0.17Now see what percentage each pair contributes to the overall effect.
contrib_Q8v9 <- 100*(chisq_Q8v9$residuals^2/chisq_Q8v9$statistic)
round(contrib_Q8v9, 3)##
## I don't like seeing them I love seeing them
## B1 <=10 0.649 1.246
## B2 11-51 0.624 0.250
## B3 52-102 0.253 0.185
## B4 103-153 2.865 0.449
## B5 154-204 0.865 0.136
## B6 206-255 5.093 0.016
## B7 256-299 19.988 0.287
## B8 >= 300 0.312 0.001
##
## Neither like nor dislike
## B1 <=10 41.555
## B2 11-51 10.380
## B3 52-102 6.937
## B4 103-153 4.775
## B5 154-204 1.441
## B6 206-255 1.171
## B7 256-299 0.450
## B8 >= 300 0.072Plot the relationships between pairs.
corrplot(chisq_Q8v9$residuals, is.cor = FALSE)
Run a G-test
GTest(x = Q8v9_table, correct = "none")##
## Log likelihood ratio (G-test) test of independence without correction
##
## data: Q8v9_table
## G = 29.825, X-squared df = 14, p-value = 0.008068Pairwise G-test
pairwise_Q8vQ9 <- pairwise.G.test(Q8v9_table)
pairwise_Q8vQ9##
## Pairwise comparisons using G-tests
##
## data: Q8v9_table
##
## B1 <=10 B2 11-51 B3 52-102 B4 103-153 B5 154-204 B6 206-255
## B2 11-51 0.427 - - - - -
## B3 52-102 0.038 0.117 - - - -
## B4 103-153 0.063 0.163 0.511 - - -
## B5 154-204 0.273 0.448 0.757 1.000 - -
## B6 206-255 0.273 0.418 0.757 0.418 0.577 -
## B7 256-299 0.320 0.384 0.511 0.273 0.427 0.806
## B8 >= 300 0.273 0.622 0.495 0.384 0.577 0.674
## B7 256-299
## B2 11-51 -
## B3 52-102 -
## B4 103-153 -
## B5 154-204 -
## B6 206-255 -
## B7 256-299 -
## B8 >= 300 0.520
##
## P value adjustment method: fdrrelationship v attitude for SEOT
From qualtrics: There is a statistically significant relationship between Q3: What is your relationship with Homer? and Q9_4: What is your general attitude towards sea otters and harbor seals? | sea otters
Q3v9 = att %>% select ("Q3", "Q9_1", "ResponseId")
dim(Q3v9)## [1] 331 3remove any rows with NAs
Q3v9 <- na.omit(Q3v9)
dim(Q3v9)## [1] 331 3Response IDs R_1rue2tn8fn8a5Ps R_3XE5rI11Shr6nYZ R_3SH2f5YqJ1todmk R_7LdtSOSwT5bECw1 R_12KtQUffRaYUb7b
still had no value for q9 so I omitted them.
Q3v9 = Q3v9[!grepl("R_1rue2tn8fn8a5Ps", Q3v9$ResponseId),]
Q3v9 = Q3v9[!grepl("R_3XE5rI11Shr6nYZ", Q3v9$ResponseId),]
Q3v9 = Q3v9[!grepl("R_3SH2f5YqJ1todmk", Q3v9$ResponseId),]
Q3v9 = Q3v9[!grepl("R_7LdtSOSwT5bECw1", Q3v9$ResponseId),]
Q3v9 = Q3v9[!grepl("R_12KtQUffRaYUb7b", Q3v9$ResponseId),]
dim(Q3v9)## [1] 326 3Response IDs R_1ZZ5435aHC9X0Ax R_3w0Qkvfaatk3Lyr
still had no value for Q3 so I omitted them.
Q3v9 = Q3v9[!grepl("R_1ZZ5435aHC9X0Ax", Q3v9$ResponseId),]
Q3v9 = Q3v9[!grepl("R_3w0Qkvfaatk3Lyr", Q3v9$ResponseId),]
dim(Q3v9)## [1] 324 3Create a contingency table (I made sure it matches what was in Qualtrics)
Q3v9_table = table(Q3v9$Q3, Q3v9$Q9_1)
#Q3v9_tableRun a Chi-square test for independence; these results are the same as what came from Qualtrics
chisq_Q3v9 = chisq.test(Q3v9_table)
chisq_Q3v9##
## Pearson's Chi-squared test
##
## data: Q3v9_table
## X-squared = 36.846, df = 6, p-value = 1.887e-06Look at which pairs contribute most to the significance. First look at the Pearson’s residuals
round(chisq_Q3v9$residuals, 2)##
## I don't like seeing them
## I am not an Alaska resident -0.62
## I am visiting from my home in Alaska 0.81
## I live here 0.18
## I work here -0.43
##
## I love seeing them
## I am not an Alaska resident 0.46
## I am visiting from my home in Alaska 0.14
## I live here -1.02
## I work here -0.15
##
## Neither like nor dislike
## I am not an Alaska resident -1.96
## I am visiting from my home in Alaska -1.32
## I live here 5.24
## I work here 1.13Now see what percentage each pair contributes to the overall effect.
contrib_Q3v9 <- 100*(chisq_Q3v9$residuals^2/chisq_Q3v9$statistic)
round(contrib_Q3v9, 3)##
## I don't like seeing them
## I am not an Alaska resident 1.048
## I am visiting from my home in Alaska 1.774
## I live here 0.090
## I work here 0.503
##
## I love seeing them
## I am not an Alaska resident 0.570
## I am visiting from my home in Alaska 0.051
## I live here 2.805
## I work here 0.065
##
## Neither like nor dislike
## I am not an Alaska resident 10.438
## I am visiting from my home in Alaska 4.754
## I live here 74.414
## I work here 3.487Plot the relationships between pairs.
corrplot(chisq_Q3v9$residuals, is.cor = FALSE)
Run a G-test
GTest(x = Q3v9_table, correct = "none")##
## Log likelihood ratio (G-test) test of independence without correction
##
## data: Q3v9_table
## G = 25.435, X-squared df = 6, p-value = 0.0002835Pairwise G-test
pairwise_Q3vQ9 <- pairwise.G.test(Q3v9_table)
pairwise_Q3vQ9##
## Pairwise comparisons using G-tests
##
## data: Q3v9_table
##
## I am not an Alaska resident
## I am visiting from my home in Alaska 0.65700
## I live here 0.00021
## I work here 0.36197
## I am visiting from my home in Alaska
## I am visiting from my home in Alaska -
## I live here 0.00228
## I work here 0.36197
## I live here
## I am visiting from my home in Alaska -
## I live here -
## I work here 0.65700
##
## P value adjustment method: fdr